Have you ever stood in front of the waste bins at the DUC or in the library and wondered to yourself, “Does this item go into the landfill, recycling, or compost bin?” We’ve all been there. Classifying wastes into those three bins can get pretty hard – especially when there are exceptions, and worse still, exceptions to those exceptions. Many strategies have been implemented to address the issue of bin contamination (wastes that are present in certain bins that really shouldn’t be) involving the use of audiovisual media, notices, and more. These strategies have varying degrees of effectiveness because they do not address the largest variable in the process – humans.
In an effort to increase sustainability on campus, this project strives to reduce human error when identifying and classifying wastes into their respective bins by utilizing computer vision.
Computer vision uses techniques from machine learning – algorithms and mathematical ideas that allow computers to learn – to deal with the task of understanding images.
Did you know that if the presence of contaminants in a load of recyclables is too high, the recyclable bags have a high chance of being sent to the landfill? This is especially true if the waste management company does not hire workers to sort out the wastes.
This project comes in two stages:
- Part I: In collaboration with the Data Science department, the library will work with students from the data science program to develop AI that can classify the most common wastes we see on campus. This will be a project for course credit.
- Part II: In collaboration with the Mechanical Engineering department, the library will work with students from mechanical design courses to design and manufacture a machine that will sort wastes based on input from the AI from Part I.